How to replace dendrogram labels using only base R and/or ggplot2 packages?

Question

I want to perform dendrogram visualization using hierarchical grouping with Minkowski method on my dataset from eurostat library. I want to make values shown in this dendrogram:

to display country names like in this one

I can only use base R packages and/or ggplot2 due to project's requirements.

Use this code to recreate my situation:

install.packages("eurostat")
install.packages("dplyr")
install.packages("ggplot2")
library(eurostat)
library(dplyr)
library(ggplot2)

member_states <- c("AT", "BE", "BG", "HR", "CY", "CZ",
                        "DK", "EE", "FI", "FR", "DE", "GR", 
                        "HU", "IE", "IT", "LV", "LT", "LU", 
                        "MT", "NL", "PL", "PT", "RO", "SK", 
                        "SI", "ES", "SE", "EL")

hicp <- get_eurostat("prc_hicp_manr", time_format = "date")

hicp_filtered <- hicp %>% filter(time >= as.Date("2000-02-01")
                               & time <= as.Date("2022-09-01")) %>%
                          filter(coicop == "CP00") %>%
                          filter(geo %in% member_states) %>%
                          mutate(geo = case_when(
                            geo == "AT" ~ "Austria",
                            geo == "BE" ~ "Belgium",
                            geo == "BG" ~ "Bulgaria",
                            geo == "HR" ~ "Croatia",
                            geo == "CY" ~ "Cyprus",
                            geo == "CZ" ~ "Czech Republic",
                            geo == "DK" ~ "Denmark",
                            geo == "EE" ~ "Estonia",
                            geo == "FI" ~ "Finland",
                            geo == "FR" ~ "France",
                            geo == "DE" ~ "Germany",
                            geo == "GR" ~ "Greece",
                            geo == "HU" ~ "Hungary",
                            geo == "IE" ~ "Ireland",
                            geo == "IT" ~ "Italy",
                            geo == "LV" ~ "Latvia",
                            geo == "LT" ~ "Lithuania",
                            geo == "LU" ~ "Luxembourg",
                            geo == "MT" ~ "Malta",
                            geo == "NL" ~ "Netherlands",
                            geo == "PL" ~ "Poland",
                            geo == "PT" ~ "Portugal",
                            geo == "RO" ~ "Romania",
                            geo == "SK" ~ "Slovakia",
                            geo == "SI" ~ "Slovenia",
                            geo == "ES" ~ "Spain",
                            geo == "SE" ~ "Sweden",
                            geo == "EL" ~ "Greece",
                            TRUE ~ geo
                          ))

data <- hicp_filtered[, c(3,4,5)]

data_widened <- reshape(transform(data, 
                        id = ave(seq_along(geo), geo, FUN = seq_along)), 
                        idvar = c("id", "time"), 
                        direction = "wide", timevar = "geo")

To perform that classification analysis I tried to write this code:

distance_matrix <- dist(data_widened[3:29, ], method = "minkowski", p = 1.5)
hc <- hclust(distance_matrix, method = "ward.D2")
plot(hc)

How can I replace those weird values with country names and align clusters on my plot too look like in the desired form?

Thanks in advance.

Answer 1

We are doing the same task. Write me to get in touch on TEAMS (Ester). However, it seems that the problem regards data_widened. I used

data_pivot <- unstack(data, values ~ time, keep.names = TRUE)
row.names(data_pivot ) <- unique(data$country)

And then go ahead with

data_matrix <- as.matrix(data_pivot)

# Compute distance matrix
distance_matrix <- dist(data_matrix , method = "minkowski", p = 1.5)

Answer 2

You have got the row and column indices round the wrong way, and you also need to transpose the data.

# Remove "values." from the names of each column
names(data_widened) <- gsub("values\\.", "", names(data_widened))

distance_matrix <- dist(t(data_widened[,3:29]), method = "minkowski", p = 1.5)
hc <- hclust(distance_matrix, method = "ward.D2")
plot(hc)

Answer 3

There are 2 options here. Use the labels option in the plot() function or define the row names.

Since this is a university assignment, I'll demonstrate with an abridged "mtcars" dataset.

#test data set.
data_widened <-structure(list(mpg = c(21, 21, 22.8, 21.4, 18.7, 18.1, 14.3, 24.4, 22.8, 19.2, 17.8, 16.4, 17.3, 15.2, 10.4), 
                              disp = c(160, 160, 108, 258, 360, 225, 360, 146.7, 140.8, 167.6, 167.6, 275.8, 275.8, 275.8, 472), 
                              hp = c(110, 110, 93, 110, 175, 105, 245, 62, 95, 123, 123, 180, 180, 180, 205), 
                              drat = c(3.9, 3.9, 3.85, 3.08, 3.15, 2.76, 3.21, 3.69, 3.92, 3.92, 3.92, 3.07, 3.07, 3.07, 2.93), 
                              wt = c(2.62, 2.875, 2.32, 3.215, 3.44, 3.46, 3.57, 3.19, 3.15, 3.44, 3.44, 4.07, 3.73, 3.78, 5.25), 
                              qsec = c(16.46, 17.02, 18.61, 19.44, 17.02, 20.22, 15.84, 20, 22.9, 18.3, 18.9, 17.4, 17.6, 18, 17.98), 
                              Names = c("Mazda RX4", "Mazda RX4 Wag", "Datsun 710",  "Hornet 4 Drive", "Hornet Sportabout", "Valiant", "Duster 360",                                                                   
                                        "Merc 240D", "Merc 230", "Merc 280", "Merc 280C", "Merc 450SE", "Merc 450SL", "Merc 450SLC", "Cadillac Fleetwood")), 
                         row.names = c(NA,15L), class = "data.frame")

#define the row names to match the desired labels
rownames(data_widened) <- data_widened$Names

#cluster
distance_matrix <- dist(data_widened[ , 1:6], method = "minkowski", p = 1.5)
hc <- hclust(distance_matrix, method = "ward.D2")

#plot
plot(hc)

# or this if the row names are not defined.
plot(hc, labels=data_widened$Names)